Skip to main content
. Author manuscript; available in PMC: 2012 Sep 1.
Published in final edited form as: Neurobiol Aging. 2009 Nov 18;32(9):1588–1598. doi: 10.1016/j.neurobiolaging.2009.10.013

Table 4.

Fit Statistics for Each Mediation Model for the Male Subgroup

Model fit
X2 df X2/df RMSEA CFI
Cognitive reserve construct
 Visual-spatial model 26.93 12 2.24 0.073 0.97
 Memory model 20.17 7 2.88 0.090 0.97
 a Speed/executive model 3.43 4 0.86 0.000 1.00
 Language model 62.20 25 2.49 0.080 0.94
Brain reserve construct
 Visual-spatial model 37.56 18 2.09 0.069 0.95
 Memory model 30.39 12 2.53 0.081 0.95
 Speed/executive model 18.11 7 2.59 0.083 0.96
 Language model 45.72 33 1.39 0.041 0.97
a

Note. error variance on education set to .02 in the model. RMSEA = root mean square error of approximation. Three measures of model mis-fit are provided, including Chi-square (X2), critical ratio (X2/df), and the RMSEA. Values closer to zero are indicative of better fit. Root mean square error of approximation values ≤ 0.08 are typically considered to be indicative of a good fit, and values between .08 and .10 are generally suggestive of an adequate fit. The comparative fit index (CFI) was also evaluated and the closer the values are to 1.0, the better the fit of the data to the model. Models with CFI values ≥.95 are typically considered to be good-fitting models. Note that χ2 may be inflated due to non-normality of the data.